This article has been accepted for inclusion in a future issue of this journal.

Content is final as presented, with the exception of pagination.

IEEE SYSTEMS JOURNAL 1

Voltage Variation on Distribution Networks With

Distributed Generation: Worst Case Scenario
M. A. Mahmud, Member, IEEE, M. J. Hossain, Senior Member, IEEE, and H. R. Pota

Abstract—This paper presents an analytical approach to estab- of centralized control is introduced in [5] to increase the DG
lish a relationship between the voltage variation and distributed connection capacity. A centralized distribution management
generation (DG) integration for the planning and operation of system controller, similar to that used in transmission networks,
distribution networks with DG. The proposed approach is mainly
based on the derivation of a voltage variation formula for distri- is presented in [6] and [7] to maximize the DG penetration using
bution networks with DG and the consideration of the worst case a wide-area voltage control and reactive power management
scenario, which establishes a relationship between the amount of approach where state estimations are employed to assess the
voltage variation and maximum permissible DG. Some recommen- voltage level. However, the centralized approaches presented
dations are presented based on the worst case voltage variation in [5] and [7] require significant investment in sensors and
formula and DG integration to counteract the voltage variation
effect. The relationship between the connection cost and voltage communication assets, which make their application difficult to
level is also presented in this paper. The feasibility of the proposed increase the DG penetration.
approach is validated by comparing the voltage profile obtained There are several alternative approaches, which include to
from the derived formula to that with the existing power system ensure the maximum DG capacity with minimal voltage im-
simulation software. pacts in the distribution network. A distributed voltage con-
Index Terms—Distributed generation (DG), distribution trol approach is introduced in [8] to limit the voltage rise
systems, voltage variation, worst case scenario. in distribution feeders by considering continuously distributed
loads and generation. Some other distributed active manage-
I. I NTRODUCTION ment approaches are presented in [9]–[11] to accommodate
more DG units by controlling the power factor rather than the

T RADITIONALLY, distribution networks are passive net-

works where the flow of both real and reactive power
is from a higher to a lower voltage level. The integration of
voltage control to counteract the effect of the DG integration.
However, these approaches are similar to fit-and-forget manner
that accommodate the full DG capacity without encountering
a small- and medium-sized generation into distribution net- any voltage or thermal issues [12].
works is increasing as these types of generating units offer a There is significant ongoing research interest for evaluating
number of technical, environmental, and economical benefits the DG capacity with a wide range of methods, objectives, and
for the utilities along with consumers due to their location constraints within two broad approaches. The first approach
near to customers [1]–[3]. Beside these benefits, the integration is to accommodate DG units with distributed and prespecified
of distributed generation (DG) units significantly changes the capacities at the best locations by using evolutionary computa-
behavior of the distribution network operation, e.g., passive tions such as genetic algorithms (GAs) [13], [14] and the parti-
distribution networks with a unidirectional power flow are cle swarm optimization (PSO) [15], which are able to handle the
converted into active distribution networks with a bidirectional distributed formulations. However, the distributed formulation
power flow. From the technical point of view, these changes of the DG capacity will not provide a truly optimal solution.
have negative impacts along with positive impacts to both Moreover, the use of multiple capacities extends the search
distribution network service providers (DNSPs) and customers. space significantly. The other approach requires the network
To provide the maximum benefit and alleviate the limit on locations of interest to be prespecified with algorithms guiding
generation capacity, an active network management (ANM) of the DG capacity growth within network constraints. In this case,
distribution networks or other novel approaches are needed to these methods are not capable of solving the continuous func-
be considered, which allow the connection of more DG units tions of the capacity. To solve these continuous functions, linear
into the existing distribution network [4]. An ANM in the form programming [16], gradient search [17], or optimal power flow
(OPF) [18]–[23] methods are employed. A major drawback
of this approach is that the perceived optimal solution may
Manuscript received September 30, 2011; revised June 12, 2012 and July 29,
2012; accepted October 7, 2012.
contain a number of sites with very small available capacities
M. A. Mahmud is with the Future Grid Research Centre, Department of Elec- when a large number of locations are searched. Moreover, the
trical and Electronic Engineering, The University of Melbourne, Melbourne, requirement of prespecifying locations is another major issue
Vic. 3010, Australia (e-mail: [email protected]).
M. J. Hossain is with Griffith School of Engineering, Griffith University, with this approach as this involves a significant effort beyond
Gold Coast, Qld. 4222, Australia (e-mail: [email protected]). the feasibility of manual searches even for a small distribution
H. R. Pota is with the School of Engineering and Information Technology, network. To overcome some of these limitations, a hybrid
The University of New South Wales at the Australian Defence Force Academy,
Canberra, A.C.T. 2600, Australia (e-mail: [email protected]). method (the combination of GA and OPF) is presented in [24]
Digital Object Identifier 10.1109/JSYST.2013.2265176 where the GA is used to search a large range of combinations

1932-8184/$31.00 © 2013 IEEE

 This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

2 IEEE SYSTEMS JOURNAL

of locations and an OPF is used to define the available capacity

for each combination. However, this method also requires the
number of DG units to be prespecified. Another hybrid method
(the combination of PSO and OPF) is introduced in [25] to
do the same task as presented in [24] and obtain a better
performance, but this method also suffers from the limitation Fig. 1. Conventional two-bus distribution system.
of prespecifying DG units.
The majority of the optimization techniques are complex In Fig. 1, DS and OLTC stand for the distribution substation
and involve more mathematical computations. Moreover, these and on-load tap-changer, respectively; VS is the sending-end
techniques have the problem of long convergence time, prema- voltage; VR is the receiving-end voltage; P and Q are the
ture convergence, and no performance guarantee. To overcome real and reactive power flowing through the distribution line,
these difficulties, analytical approaches can be adopted to en- respectively; and PL and QL are the real and reactive power
hance the power quality of distribution feeders in terms of the of the load, respectively. The voltage at the sending end can be
node voltage profile improvement, which, in turn, reduces the written as [32]
power loss. Various analytical methods have been presented in
[26]–[30] for the placement of optimal DG units in distribution P − jQ
VS = VR + (R + jX)
networks. Most of the analytical approaches as presented in VS∗
[26]–[30] are mainly based on the network sensitivity approach.
RP + XQ XP − RQ
From the existing literature on the DG planning and opera- = VR + +j .
tion, it can be summarized that the main purpose of integrating 
V ∗ V ∗
S S
DG units is to deliver a portion of real and/or reactive power
into the network to enhance the voltage profile of distribution Therefore, the voltage drop between the sending and receiv-
networks along with the reduction in distribution losses and ing ends can be written as
investment costs. However, most of the approaches are mainly
RP + XQ XP − RQ
based on the consideration of real power available from the DG ΔV = VS − VR = +j .

V ∗ V ∗
unit although the consideration of reactive power available from S S
the DG unit has gained attraction in some recent investigations
Since the phase difference between the sending-end voltage
[11], [29]–[31]. The reactive power from DG units can help
and the receiving-end voltage is very small, the voltage drop is
improve the voltage profile and reduce the energy loss. The lack
approximately equal to the real part of the drop [33], and if the
of attention to the reactive power capability exhibited by DG
sending-end bus is considered as the reference bus, the angle
units at the planning stage may lead to the potential increase
of this voltage is 0, i.e., VS∗ = |VS | = VS . Therefore, the above
in investment costs and improper allocation of DG units. It
equation can be approximated as
is of vital importance that the reactive capability limit of DG
units is accounted in distribution system planning and operation RP + XQ
problems to quantify the associated benefits. ΔV ≈ . (1)
VS
The worst case scenario of a system is very important for
the planning and operation as it captures all unusual conditions If the sending-end voltage of the system is considered as the
of the system. In distribution networks planning and operation, base voltage, in per unit, VS can be considered as one and (1)
the worst case scenario is essential to demonstrate the relation- can be written as follows:
ship between the voltage variation and DG connected to the
distribution network and to ensure that distribution networks ΔV ≈ RP + XQ. (2)
and customers will not be adversely affected. The aim of this
paper is to establish a relationship between the voltage variation The amount of voltage variation in a large distribution net-
and DG integration in a distribution network by considering the work can be also determined by using a similar approach. In
worst case scenario and to provide some solutions to the voltage Fig. 2, an n-bus system is considered. The voltage drop between
variation and connection costs. Finally, simulation results are the ith and jth buses can be written as
presented for the justification of the proposed approach.
Rij Pij + Xij Qij
ΔVij ≈ , i, j = 1, 2, 3, . . . , n, i = j
Vi
II. VOLTAGE VARIATION IN C ONVENTIONAL (3)
D ISTRIBUTION N ETWORK
where ΔVij is the variation of voltage between the ith and jth
Since distribution networks have high resistance to reactance buses, Rij is the resistance between the ith and jth buses, Xij
ratio, as compared to transmission networks, the percentage is the reactance between the ith and jth buses, Vi is the voltage
voltage drop in distribution networks is more than that in at the ith bus, and Pij and Qij are the active and reactive power
transmission networks. To get an understanding of the voltage flowing from the ith to the jth bus, respectively.
drop in distribution networks, the amount of the voltage drop The derivation of voltage variation formula for distribution
is first calculated for a two-bus distribution system, as shown networks with DG based on the derived voltage variation for-
in Fig. 1. mula as aforementioned is presented in the following section.
 This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

MAHMUD et al.: VOLTAGE VARIATION ON DISTRIBUTION NETWORKS WITH DG: WORST CASE SCENARIO 3

Fig. 2. Conventional n-bus large distribution system.

Fig. 4. n-bus large distribution system with DG.

(CHP) generators, induction machine-based wind generators,

and voltage source inverter-based photovoltaic (PV) generators
into distribution networks, which have different reactive power
capabilities. For example, if there is a synchronous machine-
Fig. 3. Two-bus distribution system with DG. based CHP unit within distribution systems, it can provide
reactive power to the system as it possesses the capability
III. VOLTAGE VARIATION IN D ISTRIBUTION of controlling reactive power output by adjusting the field
N ETWORK W ITH DG excitation. However, if a wind generator is connected to the
distribution network as DG, it is essential to provide some
When generators are connected to the distribution system,
reactive power to the system to maintain the voltage stability
the power flow and the voltage profile are affected. In order to
as its induction generator requires a source of reactive power to
export power, a generator is likely to operate at a higher voltage,
operate. Again, if a PV generator is connected to the system, it
as compared to other nodes where it is supplying power. In this
cannot provide any reactive power as it needs to operate at unity
case, receiving-end voltage VR will be
power factor in order to deliver maximum power into the grid.
VR ≈ VS + RP + XQ (4) The voltage variation in large distribution networks with
DG can be obtained in a similar way to that in conventional
as the direction of the power flow is reversed. Thus, the voltage distribution networks as discussed before. If we integrate a DG
at the point of connection of the generator will rise above the unit at the jth bus, as presented in Fig. 4, the voltage variation
sending-end voltage, which can be clarified through Fig. 3. ΔVji at the point of the DG connection, i.e., at the jth bus of a
In Fig. 3, a distributed energy resource (DER) is connected radial distribution feeder, can be written as
where the voltage is VGEN ; PG and QG are active and reactive
Rij (PGj − PLj ) + Xij (±QGj − QLj )
power generated by the DG, respectively; PL and QL are the ΔVji ≈ (7)
active and reactive power of the load, respectively; and QC is Vj
reactive power of the shunt compensator. This DER with loads
where PGj is the active power supplied by the DG unit con-
and compensators is connected to the distribution system via an
nected to the jth bus, QGj is the reactive power of the DG unit
overhead distribution line with impedance R + jX and through
connected to the jth bus, and PLj and QLj are the active and
an OLTC. The voltage variation along the distribution network,
reactive power of the load connected to the jth bus, respectively.
as shown in Fig. 3, can be written as follows:
Now, if a shunt compensator with reactive power QCj is con-
RP + XQ nected at the jth bus, (7) can be written as
ΔV = VGEN − VS ≈ (5)
VGEN
Rij (PGj − PLj ) + Xij (±QGj ± QCj − QLj )
ΔVji ≈ . (8)
where P = (PG − PL ), and Q = (±QC − QL ± QG ). If Vj
VGEN is considered as base voltage and expressed in terms of
per unit, then (5) can be written as In per unit, (8) can be written as

ΔV = VGEN − VS ≈ R(PG − PL ) + X(±QC − QL ± QG ). ΔVji ≈ Rij (PGj − PLj ) + Xij (±QGj ± QCj − QLj ).
(6)
These voltage variation formulas can be used to determine
DERs always export active power (+PG ) and may export the relationship between the voltage variation and DG inte-
or import reactive power (±QG ), whereas the load consumes gration, which may suggest possible solutions to mitigate the
both active −PL and reactive −QL power and the compen- voltage variation. In this paper, the worst case scenario of the
sators may supply or absorb only reactive power ±QC . The distribution network is considered to investigate the relationship
reactive power capability of DERs depends on the nature of between the voltage variation and DG integration, which is
DERs connected to distribution networks. The present trend is not considered in [32]. The worst case scenario of distribution
to integrate synchronous machine-based combined heat power networks with DG is discussed in the following section.
 This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

4 IEEE SYSTEMS JOURNAL

IV. E STIMATION OF VOLTAGE VARIATION AND supplied by DERs. Since there is a linear relationship between
DG C APACITY U SING W ORST C ASE S CENARIO the voltage variation and the amount of active power supplied
by DERs, the voltage variation is more onerous when there is
DERs are connected to the distribution system due to the
no demand on the system as the generation is exported back
technological innovations and changes in the economic and
to the primary distribution system. The voltage variation in the
regulatory environment, as well as to meet the increased load
distribution system also limits the amount of DG units that can
demand. From (6), we can write
be integrated into the distribution network, and this can be seen
VGEN − VS + RPL − X(±QC − QL ± QG ) in the following equation, which is obtained from (6):
PG ≈ (9)
R VGENmax − VS
PGmax ≈ . (12)
and for a large distribution network R
Vj − Vi + Rij PLj − Xij (±QCj − QLj ± QGj ) The capacity of the DG unit that can be accommodated in the
PGj ≈ . existing system is clearly limited by the maximum voltage at
Rij
the point of DG connection, which can be written as
From (9), it is clear that the level of DG connected to the
VGENmax − VS
distribution system depends on the voltage at the primary DS, PGmax ≤ (13)
the voltage level of the receiving end, the size of the conductors R
and the distance from the primary DS, the load demand on the and for a large distribution system, it can be written as
system, and the other generation on the system.
Vjmax − Vi
DNSPs should consider the worst case operating scenarios PGjmax ≤ .
to demonstrate the relationship between the voltage variation Rij
and the amount of DG integration so that their networks and Therefore, from the worst case scenario, it is seen that the
customers will not be adversely affected. Generally, these worst resistance of the line, as well as the voltage variation within the
case scenarios are: system, is critical for the DG integration. From the worst case
1) minimum load maximum generation; scenario of distribution networks with DG, possible solutions to
2) maximum load minimum generation; the voltage variation due to the integration of the maximum DG
3) maximum load maximum generation. unit can be obtained easily, which is presented in the following
Since the aim of this paper is to find out the maximum section.
amount of DG units that can be integrated into the distribution
network under the worst condition, the minimum load max-
imum generation scenario is considered in this paper as the V. M ITIGATION OF VOLTAGE VARIATION BASED ON
worst case scenario. Under the considered worst case scenario W ORST C ASE S CENARIO
The integration of DERs may cause an excessive voltage
PL = 0 QL = 0 PG = PGmax . variation. Traditionally, the DS is equipped with an over- or
Now, for the sake of simplicity, if we consider that the system undervoltage protection relay to protect it. The voltage protec-
is operating at unity power factor, ±QG and ±QC will be zero. tion scheme may permanently disconnect DERs, or it may even
In this case, the worst case voltage variation for a two-bus disconnect the DS from the main grid, which may cause serious
distribution network can be obtained from (6) and written in economical damage for customers and DNSPs. The voltage
the following form: variation on the DS can be mitigated through the following
approaches:
ΔVworst = VGENmax − VS ≈ RPGmax . (10) 1) by using the resistance reduction;
2) by regulating primary DS voltage VS ;
For a large system, the worst case voltage variation is 3) by using the generation curtailment;
4) by using the reactive power compensation.
ΔVworstji = Vjmax − Vi ≈ Rij PGjmax .

From (10), it is clear that the voltage variation depends on the A. Mitigation of Voltage Variation by Using the
resistance of the distribution line and the power supplied by Resistance Reduction
DERs. If the resistance of the distribution line is constant, then
we can write If the amount of the DG connected to a distribution system is
constant, from (10), we can write
ΔVworst ∝ PGmax (11)
ΔVworst ∝ R. (14)
and that for a large system is
From (14), it is seen that the worst case voltage variation that
ΔVworstji ∝ PGjmax . considers the maximum DG penetration is directly proportional
to the resistance of the line. Therefore, the voltage variation
Therefore, the voltage variation in the distribution network with in distribution networks can be reduced by decreasing the line
DG is directly proportional to the amount of active power resistance. The resistance of a line can be reduced by increasing
 This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

MAHMUD et al.: VOLTAGE VARIATION ON DISTRIBUTION NETWORKS WITH DG: WORST CASE SCENARIO 5

the conductor size for which it is essential to change the in-

frastructure of the existing distribution network infrastructure,
which is very difficult in practice. Therefore, before construct-
ing a new distribution system, i.e., for the future planning and
operation of a distribution network, DNSPs should consider the
reduced value of the line resistance to make provisions for a
Fig. 5. Mitigation of voltage variation by regulating primary DS voltage.
larger amount of DG units.
the nominal voltage to ensure that the voltage profile of the
B. Mitigation of Voltage Variation by Using the Reactive system remains within the specified voltage limit. As we know
Power Compensation ΔVworst = VGENmax − VS .
If we consider the worst case scenario with DERs and
compensators are operating at a power factor other than unity, The voltage variation can be mitigated by regulating the
(6) can be rewritten as sending-end voltage VS , i.e., the primary DS voltage. This
can easily be done by using OLTCs connected to distribution
ΔVworst ≈ RPGmax + X(±QGmax ± QC ). (15) networks, as shown in Fig. 5, from where it is seen that the
primary voltage can be regulated by using an automatic voltage
From (15), it is seen that the voltage variation can be mitigated controller. The controller senses the voltage variation between
by adjusting the reactive power of DERs and compensators. the two buses. If the voltage variation is within the permissible
As previously discussed, a DG unit may export or import limit, the controller does not work, but if the voltage varia-
power into or from the grid. When a DG unit exports power, tion exceeds the permissible limit, the controller automatically
(15) can be written as regulates the voltage and thus reduces the voltage variation.
However, in a more complex network, the value of this voltage
ΔVworst ≈ RPGmax + X(+QGmax ± QC ) (16) and the corresponding tap position of the OLTC would have to
be optimized.
and that for large systems

ΔVworstji ≈ Rij PGmaxj + Xij (+QGmaxj ± QCj ). (17) D. Mitigation of Voltage Variation by Using the
Generation Curtailment
From (16) and (17), it is seen that the second part of the right It is important to observe that the probability of the worst
side of these equations needs to be negative to mitigate the case scenario in distribution networks is generally low, and
voltage rise problem, and in this case, the compensator must hence, it may be beneficial to accommodate a larger DER and
absorb reactive power that should be greater than the maximum curtail it when the voltage at the busbar where it is connected
reactive power supplied by the generator. However, if there is a varies outside the specified limit. The effect of the generation
voltage dip within the network, the compensator should supply curtailment on the voltage variation can be obtained from the
reactive power to the system to keep the voltage within the following equation:
specified limit.
Again, if a DG unit imports reactive power, (15) can be VGENmax − VS
PGmax ≈ PGcur + . (20)
written as R
Equation (19) can be rewritten as
ΔVworst ≈ RPGmax + X(−QGmax ± QC ) (18)
ΔVworst ≈ RPGmax − RPGcur . (21)
and for large systems, it can be written as
From (21), it is seen that the voltage rise can be reduced through
ΔVworstji ≈ Rij PGmaxj + Xij (−QGmaxj ± QCj ). (19) the generation curtailment. The likelihood of the coincidence
of the minimum load with maximum generation will determine
From (18) and (19), it can be said that if there is a voltage rise the total annual energy that needs to be curtailed. As the price
within the system and the reactive power absorbed by DG is of electricity is primarily driven by the load demand and the
not sufficient, it is essential to absorb more reactive power and generation curtailment occurs typically during the period of low
the compensator should absorb reactive power to mitigate the load, the value of this energy to be curtailed is relatively low.
voltage rise problem. In case of a voltage dip, reactive power However, this approach is not suitable when there is a voltage
needs to be supplied from the compensator to compensate the drop within the system.
reactive power absorbed by the DG unit.
VI. VOLTAGE L EVEL AND DG C ONNECTION C OST
C. Mitigation of Voltage Variation by Regulating the Primary
From the analysis of the worst case voltage variation, it is
DS Voltage VS
seen that the voltage level at the point of the DG connection is
In conventional passive distribution networks, it is a common very important as it has a great impact on the overall profitabil-
practice for DNSPs to maintain the primary DS voltage above ity of both consumers and DNSPs. If a DG unit is connected to a
 This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

6 IEEE SYSTEMS JOURNAL

TABLE I
VOLTAGE L EVEL AND DG C ONNECTION C OST

higher voltage level, there is less chance of the voltage variation

within the network close to the customers. However, if a DG
unit is connected to the vicinity of the customer, the voltage
variation due to the DG integration may affect the customer’s
appliances. Therefore, the voltage level and the effect of voltage
variation have an inverse relationship.
The connection cost of a device into the existing distribution Fig. 6. Fifteen-bus Kumamoto, Japanese distribution system.
network is also very important. These costs are determined
based on the network access charge (NAC). Energy Australia
defines the NAC as a fixed charge (cents/day) applied to each
energized connection point at which Energy Australia’s energy/
demand is measured or recorded. Energy Australia’s NAC
under different voltage levels and network categories is shown
in Table I [34].
In Table I, the voltage levels are categorized as follows: low
voltage (LV): nominally 240/415 V; high voltage (HV): nomi-
nally 5, 6.6, 11, or 22 kV; and subtransmission (ST): nominally
33 kV or above. From Table I, it is seen that the NAC is more for
the higher voltage level, which indicates the higher connection
cost. Due to this reason, DNSPs’ main target is to connect a DG
unit into the LV level but this might cause the voltage variation
within the system. On the other hand, the connection of a DG
unit at a higher voltage level has a lower impact on the perfor-
mance of the network in terms of the steady-state voltage profile
and power quality. These two conflicting objectives need to be Fig. 7. Voltage profile of the 15-bus distribution system without DG: (solid
balanced appropriately. line with star) PSSE; (solid line with circle) derived formula; and (two dotted
lines) allowable range of voltage variation.

VII. S IMULATION R ESULTS

The voltage level at each connection point of the load and
generation is very important for the quality of the supply.
Since there are no internationally agreed rules that define the
allowed steady-state voltage range, the maximum permitted
voltage variation on each busbar is defined by some technical
regulations or specific contracts. DNSPs should maintain the
voltage variation in distribution networks within the permissible
limit fixed by national and international standards to guarantee
reliable and economic service to customers. In most cases, the
allowable voltage variation along distribution networks is ±6%
[35] but it may vary depending on rules and regulations of
each country. In this paper, the allowed voltage variation at the
customer end is considered as ±6% in order to demonstrate
the effect of the DG integration into distribution networks
with the derived voltage variation formula. In order to validate Fig. 8. Voltage profile of the 15-bus distribution system with DG: (solid line
with star) PSSE; (solid line with circle) derived formula; and (two dotted lines)
the derived voltage variation formula, the voltage profile of a allowable range of voltage variation.
distribution network is obtained from the derived formula and
compared with that of the existing power system simulation of this distribution system is 10 MVA and the base voltage is
software. 6.6 kV; the total load on the system is 6.301 MW, 0.446 MVAr.
In this paper, a 15-bus Kumamoto, Japanese distribution The test system data with the distribution of loads in different
system is considered, which is shown in Fig. 6. The base power nodes can be seen in [36].
 This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

MAHMUD et al.: VOLTAGE VARIATION ON DISTRIBUTION NETWORKS WITH DG: WORST CASE SCENARIO 7

The voltage profile of the 15-bus test system is shown in [3] H. A. Gill and G. Joos, “Models for quantifying the economic benefits
Fig. 7 from where it is seen that the voltages at buses 6, 15, of distributed generation,” IEEE Trans. Power Syst., vol. 23, no. 2,
pp. 327–335, May 2008.
and 11 are below the considered specified limit ±6% because [4] J. A. P. Lopes, N. Hatziargyriou, J. Mutale, P. Djapic, and N. Jenkins,
these buses are far away from the main grid supply point. At “Integrating distributed generation into electric power systems: A review
this point, no DERs are connected to the distribution network. of drivers, challenges and opportunities,” Elect. Power Syst. Res., vol. 77,
no. 9, pp. 1189–1203, Jul. 2007.
The voltage profile obtained from a widely used simulation tool [5] S. Liew and G. Strbac, “Maximising penetration of wind generation in
in power industry, called PSSE, is shown by the solid line with existing distribution networks,” Proc. Inst. Elect. Eng.—Gener., Transm.
stars in Fig. 7. The voltage profile of this test system should Distrib., vol. 149, no. 3, pp. 256–262, May 2002.
[6] A. Shafiu, T. Bopp, I. Chilvers, and G. Strbac, “Active management and
be similar or close to that of obtained through PSSE when the protection of distribution networks with distributed generation,” in Proc.
derived voltage variation formula will be used. With the derived IEEE Power Eng. Soc. Gen. Meet., Jun. 2004, pp. 1098–1103.
voltage variation formula, the voltage profile of the 15-bus test [7] T. Bopp, A. Shafiu, I. Cobelo, I. C. Jenkins, N. Jenkins, G. Strbac,
H. Li, and P. Crossley, “Commercial and technical integration of dis-
system is also shown in Fig. 7 by the solid line with circles, tributed generation into distribution networks,” presented at the 17th Int.
which is very close that of obtained from PSSE. Conf. Electricity Distribution (CIRED), Barcelona, Spain, 2003, Paper 45.
Now, if a 6-kW synchronous generator is connected at bus [8] M. H. J. Bollen and A. Sannino, “Voltage control with inverter-based
distributed generation,” IEEE Trans. Power Del., vol. 20, no. 1, pp. 519–
4, a 2-kW PV generator is connected at bus 13, and a 3-kW 520, Jan. 2005.
wind generator is connected at bus 10, the voltage profile of [9] C. M. Hird, H. Leite, N. Jenkins, and H. Li, “Network voltage controller
the distribution network is shown in Fig. 8. In this case, there for distributed generation,” Proc. Inst. Elect. Eng.—Gener., Transm. Dis-
trib., vol. 151, no. 2, pp. 150–156, Mar. 2004.
should be a voltage rise, as compared with the distribution [10] A. E. Kiprakis and A. R. Wallace, “Maximizing energy capture from
network without DG at buses 4 and 13, as well as their adjacent distributed generators in weak networks,” Proc. Inst. Elect. Eng.—Gener.,
buses and a voltage drop at bus 10 and its adjacent buses. This Transm. Distrib., vol. 151, no. 5, pp. 611–618, Sep. 2004.
[11] P. M. S. Carvalho, P. F. Correia, and L. A. F. M. Ferreira, “Distributed
happens as the synchronous generator supplies both active and reactive power generation control for voltage rise mitigation in distribu-
reactive power and the PV generator supplies active power to tion networks,” IEEE Trans. Power Syst., vol. 23, no. 2, pp. 766–772,
the system. Moreover, this portion of the system has enough May 2008.
[12] A. Keane, L. F. Ochoa, E. Vittal, C. J. Dent, and G. P. Harrison, “Enhanced
reactive power to supply loads. On the other hand, the region at utilization of voltage control resources with distributed generation,” IEEE
which the wind generator is connected is far from the main grid, Trans. Power Syst., vol. 26, no. 1, pp. 252–260, Feb. 2011.
and the reactive power available in this portion is not sufficient [13] G. Celli, E. Ghiani, S. Mocci, and F. Pilo, “A multiobjective evolutionary
algorithm for the sizing and siting of distributed generation,” IEEE Trans.
to meet the load demand. Moreover, the wind generator also Power Syst., vol. 20, no. 2, pp. 750–757, May 2005.
consumes reactive power, and this is why the voltage at bus 11, [14] L. F. Ochoa, A. Padilha-Feltrin, and G. P. Harrison, “Time-series-based
which is adjacent to bus 10, falls down outside the specified maximization of distributed wind power generation integration,” IEEE
Trans. Energy Convers., vol. 23, no. 3, pp. 968–974, Sep. 2008.
limit. In Fig. 8, it is shown that the voltage profile obtained [15] A. M. El-Zonkoly, “Optimal placement of multi-distributed generation
from the derived formula is very close to that of obtained from units including different load models using particle swarm optimisation,”
PSSE when DERs are integrated into the distribution system. IET Gener., Transm. Distrib., vol. 5, no. 7, pp. 760–771, Jul. 2011.
[16] A. Keane and M. OMalley, “Optimal allocation of embedded genera-
tion on distribution networks,” IEEE Trans. Power Syst., vol. 20, no. 3,
pp. 1640–1646, Aug. 2005.
VIII. C ONCLUSION [17] N. S. Rau and Y. H. Wan, “Optimum location of resources in dis-
tributed planning,” IEEE Trans. Power Syst., vol. 9, no. 4, pp. 2014–2020,
The voltage variation for a small and a large distribution Nov. 1994.
[18] G. P. Harrison and A. R. Wallace, “Optimal power flow evaluation of dis-
network has been estimated through an analytical approach, and tribution network capacity for the connection of distributed generation,”
the derived voltage variation formula is accurate for performing Proc. Inst. Elect. Eng.—Gener., Transm. Distrib., vol. 152, no. 1, pp. 115–
the analysis as it provides very similar results to that of the 122, Jan. 2005.
[19] P. N. Vovos, G. Harrison, A. R. Wallace, and J. W. Bialek, “Optimal power
existing power system simulation software. The worst case flow as a tool for fault level constrained network capacity analysis,” IEEE
scenario of the distribution network is considered based on the Trans. Power Syst., vol. 20, no. 2, pp. 734–741, May 2005.
derived formula in order to build up a relationship between the [20] P. N. Vovos and J. W. Bialek, “Direct incorporation of fault level con-
straints in optimal power flow as a tool for network capacity analysis,”
voltage variation and maximum DG capacity. The relationship IEEE Trans. Power Syst., vol. 20, no. 4, pp. 2125–2134, Nov. 2005.
among voltage level, voltage variation, and DG connection cost [21] L. F. Ochoa, C. J. Dent, and G. P. Harrison, “Distribution network capacity
is built using an example. The approach presented in this paper assessment: Variable DG and active networks,” IEEE Trans. Power Syst.,
vol. 25, no. 1, pp. 87–95, Feb. 2010.
is useful for the future planning and operation of distribution [22] C. J. Dent, L. F. Ochoa, G. P. Harrison, and J. W. Bialek, “Efficient secure
networks, as well as for the voltage control of the existing AC OPF for network generation capacity assessment,” IEEE Trans. Power
distribution network with DG. Future works will deal with the Syst., vol. 25, no. 1, pp. 575–583, Feb. 2010.
[23] C. J. Dent, L. F. Ochoa, and G. P. Harrison, “Network distributed gen-
detailed analysis of the voltage control with different types of eration capacity analysis using OPF with voltage step constraints,” IEEE
DG such as CHP, wind generators, and PV. Trans. Power Syst., vol. 25, no. 1, pp. 296–304, Feb. 2010.
[24] G. P. Harrisona, A. Piccolo, P. Siano, and A. R. Wallacen, “Hybrid
GA and OPF evaluation of network capacity for distributed genera-
R EFERENCES tion connections,” Elect. Power Syst. Res., vol. 78, no. 3, pp. 392–398,
Mar. 2008.
[1] P. Chiradeja and R. Ramakumar, “An approach to quantify the technical [25] M. Gomez-Gonzaleza, A. Lopez, and F. Jurado, “Optimization of dis-
benefits of distributed generation,” IEEE Trans. Energy Convers., vol. 19, tributed generation systems using a new discrete PSO and OPF,” Elect.
no. 4, pp. 764–773, Dec. 2004. Power Syst. Res., vol. 84, no. 1, pp. 174–180, Mar. 2012.
[2] A. G. Tsikalakis and N. D. Hatziagyriou, “Environmental benefits of [26] C. Wang and M. H. Nehrir, “Analytical approaches for optimal placement
distributed generation with and without emissions trading,” Energy Policy, of distributed generation sources in power systems,” IEEE Trans. Power
vol. 35, no. 6, pp. 3395–3409, Jun. 2007. Syst., vol. 19, no. 4, pp. 2068–2076, Nov. 2004.
 This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

8 IEEE SYSTEMS JOURNAL

[27] N. Acharya, P. Mahat, and N. Mithulananthan, “An analytical approach M. J. Hossain (M’11–SM’13) received the B.Sc.
for DG allocation in primary distribution network,” Int. J. Elect. Power and M.Sc.Eng. degrees from Rajshahi University
Energy Syst., vol. 28, no. 10, pp. 669–678, Dec. 2006. of Engineering and Technology (RUET), Rajshahi,
[28] T. Gozel and M. H. Hocaoglu, “An analytical method for the sizing and Bangladesh, in 2001 and 2005, respectively, and the
siting of distributed generators in radial systems,” Elect. Power Syst. Res., Ph.D. degree from The University of New South
vol. 79, no. 6, pp. 912–918, Jun. 2009. Wales, Canberra, Australia, in 2010, all in electrical
[29] D. Q. Hung, N. Mithulananthan, and R. C. Bansal, “Analytical expressions and electronic engineering.
for DG allocation in primary distribution networks,” IEEE Trans. Energy He served as a Research Fellow in the School of
Convers., vol. 25, no. 3, pp. 814–820, Sep. 2010. Information Technology and Electrical Engineering,
[30] D. Q. Hung, N. Mithulananthan, and R. Bansal, “Multiple distributed The University of Queensland, Brisbane, Australia.
generators placement in primary distribution networks for loss reduction,” He was also an Assistant Professor and a Lecturer
IEEE Trans. Ind. Electron, vol. 60, no. 4, pp. 1700–1708, Apr. 2013. at RUET for six years. He is currently a Lecturer in Griffith School of
[31] K. Zou, A. P. Agalgaonkar, K. M. Muttaqi, and S. Perera, “Distribution Engineering, Griffith University, Gold Coast, Australia. His research interests
system planning with incorporating DG reactive capability and system are power systems, renewable energy integration and stabilization, voltage
uncertainties,” IEEE Trans. Sustain. Energy, vol. 3, no. 1, pp. 112–123, stability, microgrids, robust control, electrical machine, flexible alternating
Jan. 2012. current transmission system devices, and energy storage systems.
[32] M. L. Doumbia and K. Agbossou, “Voltage variation analysis in inter-
connected electrical network—Distributed generation,” in Proc. IEEE
Canada Elect. Power Conf., 2007, pp. 525–530.
[33] W. H. Kersting, Distribution System Modeling and Analysis, 2nd ed.
London, U.K.: CRC Press, 2007.
[34] “Report ‘Network pricing list’,” Energy Australia, Melbourne, VIC,
Australia, 2010.
[35] C. L. Masters, “Voltage rise: The big issue when connecting embedded
generation to long 11 kV overhead lines,” Power Eng. J., vol. 16, no. 1,
pp. 5–12, Feb. 2002.
[36] F. M. Gonzalez-Longatt, “Impact of distributed generation over power
losses on distribution system,” in Proc. 9th Int. Conf. Elect. Power Qual.
Utilisation, Oct. 2007, pp. 1–6.
H. R. Pota received the B.E. degree from Sardar
Vallabhbhai Regional College of Engineering and
M. A. Mahmud (S’10–M’13) received the Bache- Technology, Surat, India, in 1979; the M.E. degree
lor’s degree (with honours, first class first position) in from the Indian Institute of Science, Bangalore,
electrical and electronic engineering from Rajshahi India, in 1981; and the Ph.D. degree from The Uni-
University of Engineering and Technology (RUET), versity of Newcastle, Callaghan, Australia, in 1985,
Rajshahi, Bangladesh, in 2008 and the Ph.D. de- all in electrical engineering.
gree (with best thesis award) in electrical engi- He is currently an Associate Professor at The
neering from The University of New South Wales, University of New South Wales at the Australian
Canberra, Australia, in 2012. Defence Force Academy, Canberra, Australia. He
He served as a Radio Network Planning Engineer has held visiting appointments at the University of
at Huawei Technologies (BD) Ltd. After that, he be- Delaware, Newark, DE, USA; Iowa State University, Ames, IA, USA; Kansas
came a Lecturer at Khulna University of Engineering State University, Manhattan, KS, USA; Old Dominion University, Norfolk, VA,
and Technology, Khulna, Bangladesh, and in the Department of Electrical USA; the University of California at San Diego, La Jolla, CA, USA; and the
and Electronic Engineering, RUET. He is currently a Research Fellow at Centre for Artificial Intelligence and Robotics, Bangalore. He has a continuing
The University of Melbourne, Melbourne, Australia. His research interests are interest in the area of power system dynamics and control and modeling and
dynamic stability of power systems, renewable energy integration, smart grids, control of mechanical systems such as flexible structures, acoustical systems,
nonlinear control theory, and electrical machine. and unmanned aerial vehicles.